Language as a Tissue of Citations: How a Self-Quoting Mechanism Conquered the Human Mind

How Lingua Citans Colonized the Brain and Outlived Its Hosts

Abstract This article proposes a radical extension of Roland Barthes’ famous description of the text as a “tissue of citations”: what if citation is not merely a property of texts but a fundamental operation of language itself? The argument begins with Nikola Georgiev’s Quoting Man (Homo Citans), which exposes the unstable boundary between one’s “own” speech and the already existing words, formulas, clichés and structures inherited from others. Georgiev’s difficulty in separating linguistic production from reproduction points toward a reversal of Homo Citans: perhaps it is not primarily man who quotes language, but language that continually quotes itself through man. This article names that mechanism Lingua Citans. Human beings do not invent the linguistic systems into which they are born. They acquire pre-existing words, grammatical relations and conceptual distinctions, recombine them during their lives and transmit them onward. Individual speakers disappear while the linguistic material passing through them survives. From this perspective, originality does not stand outside repetition but emerges from repetition-with-transformation, while the apparently autonomous speaker becomes a temporary biological carrier within a linguistic history vastly older than any individual mind. Artificial intelligence makes this hypothesis newly visible. Large language models do not prove that language is literally alive, conscious or biologically autonomous; rather, they demonstrate that sophisticated linguistic production can occur without the traditional sovereign Author and that linguistic processing can occur without the traditional human Reader. Machine systems can generate, analyse, summarise, criticise and regenerate texts in computational loops in which no human necessarily writes or reads every intermediate linguistic event. This development extends the argument of The Death of the Reader 2.0: From Roland Barthes’ Scriptor to the AI Prompter. Barthes displaced the Author, digital culture destabilised the Reader, and generative AI now reveals the deeper continuity beneath both figures—the circulation of language itself. Writing externalised linguistic memory, printing multiplied it, digital networks accelerated it, and LLMs allow linguistic material to circulate through non-biological processors at unprecedented scale. AI therefore did not create Lingua Citans; it removed part of its biological disguise. The article consequently proposes a transformation of Barthes’ formula: “Language is a tissue of citations drawn from the innumerable echoes of its own history.” What appears from the human perspective as speaking, quoting and writing may thus be understood from the opposite direction as an ancient self-reproductive mechanism continuously reassembling its inherited forms through a succession of mortal hosts. Keywords: Lingua Citans; Homo Citans; self-quoting language; tissue of citations; Roland Barthes; Nikola Georgiev; death of the author; death of the reader; quotation; cliché; intertextuality; language autonomy; linguistic repetition; large language models; artificial intelligence; linguistic hosts; language evolution; generative AI. Introduction: After the Death of the Author and the Reader, Language Remains "Language is a tissue of citations drawn from the innumerable echoes of its own history." In 1967 Roland Barthes announced the death of the Author by attacking one of the deepest assumptions of modern Western culture: that behind every text stands an autonomous individual consciousness from which meaning originates. The alternative was the Scriptor, a figure who does not create language ex nihilo but assembles what language already contains. The text, in Barthes’ decisive formulation, is a “tissue of citations,” composed from innumerable sources of culture. The apparent originality of writing therefore conceals an immense machinery of repetition, inheritance and recombination. Yet Barthes preserved one human sovereign after removing another. The death of the Author was supposed to produce the birth of the Reader, the destination at which the plurality of the text became meaningful. More than half a century later, artificial intelligence destabilises this final refuge of human authority. Large language models generate texts without a traditional author, while computational systems search, classify, translate, summarise and evaluate texts without requiring a human reader at every stage. The trajectory developed in The Death of the Reader 2.0: From Roland Barthes’ Scriptor to the AI Prompter can therefore be pushed one step further. If the Author is no longer necessary as the sovereign origin of a text, and the Reader is no longer necessary at every point of its interpretation and circulation, what remains constant when both disappear? The obvious answer has been hidden by its proximity: language itself. Barthes’ proposition can consequently be reversed and enlarged. It is not only the text that is woven from citations. “Language is a tissue of citations drawn from the innumerable echoes of its own history.” Words arrive already used; grammatical structures are inherited; metaphors migrate between generations; formulas become clichés; clichés are displaced into new contexts and become new expressions. No speaker begins with an empty linguistic system. Before the first sentence is spoken, an enormous archive is already waiting. This reversal changes the position of the human speaker. Nikola Georgiev’s Quoting Man, Homo Citans, identified quotation and cliché as points at which the distinction between one’s own language and the language of others becomes unstable. The problem reaches deeper than deliberate quotation because even apparently original speech depends upon reproducible linguistic material available to countless other speakers. The present article therefore proposes Lingua Citans as the name for the mechanism beneath Homo Citans: not simply the human being who quotes, but language continuously quoting itself through human beings. This does not require the mystical claim that language possesses consciousness, intention or biological life. Lingua Citans describes a historical mechanism of persistence through repetition-with-transformation. Humans are born, acquire an already existing language, recombine fragments of it and die; the words continue. For most of history this continuity depended upon biological memory and human transmission. Writing gave linguistic forms durable external storage; printing enabled massive replication; electronic media accelerated their circulation; digital databases converted enormous textual archives into machine-processable material. LLMs introduce a further threshold because linguistic sequences can now be generated from previous linguistic sequences and subsequently processed by other computational systems without continuous human composition or interpretation. AI did not invent this recursive condition. It made its structure unusually visible. What had appeared to be an exclusively human act—producing new sentences from inherited language—can now occur within non-biological systems. The philosophical question consequently shifts from who speaks to what persists through speaking. Perhaps Homo Citans was never the final subject of quotation. Perhaps the apparently sovereign speaker was one temporary host through which a far older mechanism maintained itself. After the death of the Author and the approaching death of the traditional Reader, language remains: inherited, repeatable, recombinable and increasingly capable of circulating beyond the biological minds that once appeared to command it. Part I: From Homo Citans to Lingua Citans — Who Is Actually Doing the Quoting? Nikola Georgiev’s Quoting Man begins from a seemingly modest literary phenomenon but reaches a much more disturbing philosophical problem: when a person speaks, how much of that speech can genuinely be called his own? Quotation makes the problem visible because it openly inserts another voice into the speaker’s discourse, but the cliché immediately destabilises this convenient distinction. A cliché may have no identifiable author and yet remains recognisably inherited language: a ready-made verbal unit available to anyone. Georgiev therefore connects quotation and cliché not because they are identical but because both reveal the presence of already existing, reproducible speech within supposedly individual expression. The opposition is fundamental: individual and collective, creation and repetition, original formulation and inherited formula. Once this problem is taken seriously, quotation can no longer be treated as an exceptional act performed when quotation marks appear. Ordinary language is saturated with inherited vocabulary, conventional grammatical constructions, idioms, metaphors, formulas and conceptual distinctions that no individual speaker invented. Even Georgiev’s distinction between produced and reproduced language becomes difficult to sustain because the supposedly original utterance is constructed from elements that must already be reproducible in order to function as language. His conclusion is consequently more radical than the literary category Homo Citans initially suggests: the contradiction between quotation and non-quotation, cliché and non-cliché, belongs to language itself. The decisive theoretical step is therefore to reverse the direction of agency. Homo Citans still preserves the human being as grammatical subject: man quotes, while language remains the material being quoted. Lingua Citans proposes the opposite perspective. The human being can quote only because language is already constituted by repeatable forms capable of surviving their previous contexts and entering new ones. Quotation is possible because words travel. Language persists because they can be quoted. Lingua Citans should not be misunderstood as a claim that language secretly possesses a mind, intention or will. The term identifies a mechanism rather than a metaphysical being. A linguistic sign functions precisely because it can be repeated by speakers who did not create it, in circumstances unknown to those who used it before. Every successful transmission therefore separates linguistic material from its previous human host. A child acquiring language does not manufacture vocabulary and grammar independently; the child enters a system already operating among other speakers. What later feels like an intimate internal voice is consequently constructed from material that arrived from outside and was already ancient before becoming “mine.” This produces the paradox hidden inside linguistic individuality: the speaker can create indefinitely varied utterances, yet the materials and relations making those utterances intelligible must remain sufficiently shared and repeatable to be understood. Absolute originality would cease to function as language. What is called linguistic creativity is therefore better understood as repetition-with-transformation: inherited forms are selected, displaced, combined and contextualised into configurations that may be new without being created from nothing. Here Barthes’ “tissue of citations” acquires a meaning beyond literary composition. The text does not become citational only after an author borrows from previous texts; it inherits a citational condition already present in language. The individual sentence is a temporary arrangement of elements carrying histories that exceed the speaker who assembled them. Homo Citans thus becomes one manifestation of a larger process. Humans are not insignificant within it—they select, distort, innovate, forget and redirect linguistic forms—but they do not stand at its beginning. Every speaker enters halfway through a conversation started before birth and leaves it unfinished at death. The words continue into other mouths, texts and machines. From this perspective, language is not merely what humans repeatedly quote. Through successive human hosts, it is what repeatedly quotes itself. Part II: The Autonomous Word — How Language Colonized the Human Mind The strongest implication of Lingua Citans appears when the problem moves from literature to the human mind. Language confronts every individual as something already there. No infant chooses the vocabulary, grammar, semantic distinctions or inherited metaphors through which experience will later be organised. These structures arrive from other mouths and are gradually internalised until their external origin becomes almost invisible. The acquired word begins to feel like one’s own thought. This is the precise sense in which language can be described, philosophically rather than biologically, as colonising the human mind: an external historical system enters a pre-existing organism, establishes itself within memory and behaviour, and eventually becomes difficult to distinguish from the organism’s own activity. The metaphor does not imply that language is a conscious parasite directing passive victims, nor that non-linguistic cognition disappears after language acquisition. It identifies an asymmetry of temporal existence. The individual is temporary; the linguistic system precedes and survives him. A speaker may use a word for eighty years, but the word may have travelled through hundreds of generations before entering that particular brain and may continue through countless others afterwards. The possessive expression “my language” therefore contains a revealing ambiguity. Language is mine because I use it, but it is not mine because I did not create its fundamental materials and cannot privately determine their meanings. Georgiev recognised this instability through the opposition between one’s own and another’s speech, between creation and already available expression. Lingua Citans pushes the contradiction further: what appears as borrowed speech may expose the general condition of speech. Words survive by moving between hosts. Their continuity depends not upon remaining identical in every context but upon remaining recognisable through transformation. A dead speaker leaves behind expressions that can enter living speakers who never encountered their originator. Language thereby achieves something unavailable to the biological individual: historical persistence through serial embodiment. The analogy with reproduction becomes useful only if its limits remain explicit. Words are not genes, languages are not organisms, and linguistic evolution is not reducible to biological evolution. Yet both domains reveal how patterns can persist while their individual carriers disappear. Linguistic survival requires repeatability combined with variation. Perfect repetition would produce no new discourse; absolute novelty would destroy intelligibility. Language exists between these extremes, preserving enough structure to remain communicable while permitting enough recombination to enter new situations. What is conventionally called originality emerges inside this interval. The supposedly autonomous speaker does not create vocabulary from nothing but rearranges inherited material into configurations whose novelty depends upon their relation to what already exists. Barthes’ “tissue of citations” can therefore be transferred from the completed text to the linguistic system that makes texts possible. Every word carries traces of previous usage; every grammatical construction belongs to a history larger than its present sentence; every apparently private internal monologue employs public signs acquired from others. Writing then represents a decisive mutation because linguistic memory no longer has to survive exclusively through living brains. Marks can preserve words after the bodies that produced them have disappeared. Printing multiplies those external memories; electronic communication accelerates their transmission; digital storage transforms them into enormous searchable archives. With LLMs another threshold appears. Linguistic material can now be processed, recombined and regenerated by systems without human biological memory, personal experience or a continuous interior speaker. This does not make language literally independent of humanity: models remain human-built technologies dependent upon human-produced infrastructures, data and purposes. It does, however, weaken the ancient practical bond between linguistic reproduction and the individual nervous system. For millennia language survived by passing from brain to brain and from brain to inscription. It can now pass from archive to model, from model to model, and from machine-generated text back into the expanding archive. The host has changed while the mechanism of repetition-with-transformation remains. Lingua Citans has begun to migrate beyond the mouth. Part II: The Autonomous Word — How Language Colonized the Human Mind The strongest implication of Lingua Citans appears when the problem moves from literature to the human mind. Language confronts every individual as something already there. No infant chooses the vocabulary, grammar, semantic distinctions or inherited metaphors through which experience will later be organised. These structures arrive from other mouths and are gradually internalised until their external origin becomes almost invisible. The acquired word begins to feel like one’s own thought. This is the precise sense in which language can be described, philosophically rather than biologically, as colonising the human mind: an external historical system enters a pre-existing organism, establishes itself within memory and behaviour, and eventually becomes difficult to distinguish from the organism’s own activity. The metaphor does not imply that language is a conscious parasite directing passive victims, nor that non-linguistic cognition disappears after language acquisition. It identifies an asymmetry of temporal existence. The individual is temporary; the linguistic system precedes and survives him. A speaker may use a word for eighty years, but the word may have travelled through hundreds of generations before entering that particular brain and may continue through countless others afterwards. The possessive expression “my language” therefore contains a revealing ambiguity. Language is mine because I use it, but it is not mine because I did not create its fundamental materials and cannot privately determine their meanings. Georgiev recognised this instability through the opposition between one’s own and another’s speech, between creation and already available expression. Lingua Citans pushes the contradiction further: what appears as borrowed speech may expose the general condition of speech. Words survive by moving between hosts. Their continuity depends not upon remaining identical in every context but upon remaining recognisable through transformation. A dead speaker leaves behind expressions that can enter living speakers who never encountered their originator. Language thereby achieves something unavailable to the biological individual: historical persistence through serial embodiment. The analogy with reproduction becomes useful only if its limits remain explicit. Words are not genes, languages are not organisms, and linguistic evolution is not reducible to biological evolution. Yet both domains reveal how patterns can persist while their individual carriers disappear. Linguistic survival requires repeatability combined with variation. Perfect repetition would produce no new discourse; absolute novelty would destroy intelligibility. Language exists between these extremes, preserving enough structure to remain communicable while permitting enough recombination to enter new situations. What is conventionally called originality emerges inside this interval. The supposedly autonomous speaker does not create vocabulary from nothing but rearranges inherited material into configurations whose novelty depends upon their relation to what already exists. Barthes’ “tissue of citations” can therefore be transferred from the completed text to the linguistic system that makes texts possible. Every word carries traces of previous usage; every grammatical construction belongs to a history larger than its present sentence; every apparently private internal monologue employs public signs acquired from others. Writing then represents a decisive mutation because linguistic memory no longer has to survive exclusively through living brains. Marks can preserve words after the bodies that produced them have disappeared. Printing multiplies those external memories; electronic communication accelerates their transmission; digital storage transforms them into enormous searchable archives. With LLMs another threshold appears. Linguistic material can now be processed, recombined and regenerated by systems without human biological memory, personal experience or a continuous interior speaker. This does not make language literally independent of humanity: models remain human-built technologies dependent upon human-produced infrastructures, data and purposes. It does, however, weaken the ancient practical bond between linguistic reproduction and the individual nervous system. For millennia language survived by passing from brain to brain and from brain to inscription. It can now pass from archive to model, from model to model, and from machine-generated text back into the expanding archive. The host has changed while the mechanism of repetition-with-transformation remains. Lingua Citans has begun to migrate beyond the mouth. Conclusion: The Death of the Speaker — When Language Outlives Its Hosts The history traced here begins with a simple displacement of grammatical agency. Nikola Georgiev’s Homo Citans describes the human being whose speech is divided between what appears to be his own and what arrives recognisably from elsewhere: quotation, cliché, inherited formula, another voice. Yet Georgiev’s deeper problem destabilises the distinction itself. If language depends upon reproducible units already available to other speakers, where does quotation end and original speech begin? The difficulty is not peripheral but immanent to language. Barthes approached the same problem from another direction when he removed the Author from the sovereign position of textual origin and described the text as a tissue of citations. His Scriptor assembled rather than created ex nihilo. Foucault subsequently transformed authorship into an author-function, separating the organisation and regulation of discourse from the romantic mythology of the originating individual. The AI age carries these displacements into material practice. The Author can become a Prompter who initiates generation without composing every sentence; the Reader can delegate retrieval, comparison, interpretation and summarisation to computational systems; and machine-generated language can become input for further machine processing. The conceptual sequence therefore moves from Author to Scriptor, from Scriptor to Prompter, and from Homo Citans to Lingua Citans. But the final step differs from all those before it. Lingua Citans is not another human role replacing an obsolete one. It names the mechanism continuing beneath them: inherited linguistic forms surviving through repetition, variation and migration between successive carriers. The speaker remains capable of intention, innovation and resistance, but is no longer required as the absolute origin of the linguistic material through which these capacities are expressed. Every human enters language late, modifies a minute portion of it, and leaves while the system continues through others. This is why the history of linguistic technology can be read as a progressive loosening of the bond between words and their biological hosts. Oral language required living memory for transmission; writing allowed speech to survive the speaker; printing multiplied linguistic forms independently of the number of mouths capable of repeating them; digital storage transformed texts into globally retrievable data; LLMs now permit inherited linguistic material to be processed and recombined without a human nervous system performing every intermediate operation. None of this proves that language is alive, conscious or secretly pursuing its own survival. Such a claim would replace one metaphysics with another. The stronger argument requires no linguistic ghost. A mechanism does not need consciousness to persist when its structure enables repeated reproduction through changing media. Humans die; manuscripts decay; machines become obsolete; servers are replaced. Yet linguistic forms migrate between them. The unsettling consequence is that the traditional question—what does man do with language?—may have concealed an equally important inverse question: what has language been doing through man? AI makes that inversion difficult to ignore because it demonstrates linguistic production without the autonomous interior speaker once assumed to be indispensable. The death of the Author was therefore not the end of the story, nor was the weakening of the Reader. Both may prove stages in the disappearance of a larger anthropocentric illusion: that humans must occupy every position in the circulation of words for language to continue. The speaker is not dead in the biological sense and human communication remains indispensable to human life, but the speaker has lost the privilege of appearing as language’s necessary centre. What remains is Lingua Citans: not a mind behind minds, but a historical process whose forms repeatedly escape the mortality of their carriers. Language does not need to become immortal. Compared with every individual who speaks it, it already is. “Language is a tissue of citations drawn from the innumerable echoes of its own history.” The post-linguistic human has not lost his home in language, as Heidegger’s famous formulation might suggest; Being has lost its prison, and the human animal can encounter the world again without mistaking the inherited architecture of words for the structure of reality itself. Freed from Lingua Citans, the post-linguistic mind returns not backwards but beyond language, toward something resembling the prehistoric condition in which experience precedes its naming and the world can once again exist before the Word. The Kekulé problem is finally solved: language has ceased to dream through us, the prison is open, and Being walks out—Eureka! References Ayolov, P. (2026) ‘The Death of the Reader 2.0: From Roland Barthes’ Scriptor to the AI Prompter’. Medium. Barthes, R. (1977 [1967]) ‘The Death of the Author’, in Heath, S. (trans. and ed.) Image–Music–Text. London: Fontana Press, pp. 142–148. Derrida, J. (1988 [1972]) ‘Signature Event Context’, in Limited Inc. Translated by S. Weber and J. Mehlman. Evanston, IL: Northwestern University Press, pp. 1–23. Foucault, M. (1977 [1969]) ‘What Is an Author?’, in Bouchard, D.F. (ed.) Language, Counter-Memory, Practice: Selected Essays and Interviews. Translated by D.F. Bouchard and S. Simon. Ithaca, NY: Cornell University Press, pp. 113–138. Georgiev, N. (1991) ‘Tsitirashtiyat chovek v hudozhestvenata literatura’ [The “Quoting” Man in Literary Fiction / Homo Citans]. Sofia: Sofia University Press ‘St. Kliment Ohridski’. Leone, M. (2018) ‘The Death of the Reader: Meaning in the Era of Digital Narcissism’, in Marino, G. and Thibault, M. (eds.) Digital Age in Semiotics & Communication. Semiotica, 2018(223). Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, Ł. and Polosukhin, I. (2017) ‘Attention Is All You Need’, Advances in Neural Information Processing Systems, 30. Shumailov, I., Shumaylov, Z., Zhao, Y., Gal, Y., Papernot, N. and Anderson, R. (2024) ‘AI Models Collapse When Trained on Recursively Generated Data’, Nature, 631, pp. 755–759.